Re: [LONG] Re: LDAP Server backups

[Ron Chmara <ron@Opus1.COM>]

wesley.craig@umich.edu wrote:
> > 1 master server, 2 public "slaves", one private slave. All Redhat Linux,
> > various versions. The slaves all have backup scripts, as well as being
> > "live" backups of the master. (Can you tell I once lost data and went down for
> > 4 hours on a 20,000 user system? :-) )
> I have a question and a comments.  First, does killproc on linux wait
> for the process to die?

On RH 6 _and 7, yes, for a little bit. Not a lot, though.... if you're
running slapd with a fair bit of lag in your buffers, or with db syncing off,
or off of a really slow disk, it's not necesarrily good thing to run. 

> I ask because slapd can take a long time to
> write out it's data

You certianly don't want to be killing anything lightly. :-) That's why
I always run my killproc restart scripts in an inactive time period.

> and you shouldn't run ldbmcat on a live database.
> Second, our "private" slave also thinks it is a master, and so creates
> a relog.  We have a cron job that saves the replog each hour for a
> week.  With a weeks worth of ldifs and relogs, we can recreate the
> database at any point in the week.  This offers us protection from
> (possibly malicious) non-crash data corruption.

Wow. Great technique (hadn't even considered it, though it
makes perfect sense).

-Ronabop


--2D426F70|759328624|00101101010000100110111101110000
Personal:  ron@opus1.com, 520-326-6109, http://www.opus1.com/ron/
Work: rchmara@pnsinc.com, 520-546-8993, http://www.pnsinc.com/
The opinions expressed in this email are not necessarily those of myself,
my employers, or any of the other little voices in my head.